Proper study design is one of the most important aspects of successful biomarker discovery and that includes both a thorough understanding of the disease population and which statistical tools to use. New research from Arizona State University examines how the predictive reliability of statistical tests relies not only on the choice of the test (clearly) but the disease heterogeneity. Studies investigating diseases like cancer, that can be divided into subgroups based on molecular characteristics, may require different statistical tests than monotypic diseases. Biomarker can be used to effectively define these subgroups but consideration must be given to the statistical tools being used.